Benjamin Peterson | 90f5ba5 | 2010-03-11 22:53:45 +0000 | [diff] [blame] | 1 | #! /usr/bin/env python3 |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 2 | |
| 3 | # Convert MH directories (1 message per file) or MMDF mailboxes (4x^A |
| 4 | # delimited) to unix mailbox (From ... delimited) on stdout. |
| 5 | # If -f is given, files contain one message per file (e.g. MH messages) |
| 6 | |
| 7 | import rfc822 |
| 8 | import sys |
| 9 | import time |
| 10 | import os |
| 11 | import stat |
| 12 | import getopt |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 13 | import re |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 14 | |
| 15 | def main(): |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 16 | dofile = mmdf |
| 17 | try: |
| 18 | opts, args = getopt.getopt(sys.argv[1:], 'f') |
Guido van Rossum | b940e11 | 2007-01-10 16:19:56 +0000 | [diff] [blame] | 19 | except getopt.error as msg: |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 20 | sys.stderr.write('%s\n' % msg) |
| 21 | sys.exit(2) |
| 22 | for o, a in opts: |
| 23 | if o == '-f': |
| 24 | dofile = message |
| 25 | if not args: |
| 26 | args = ['-'] |
| 27 | sts = 0 |
| 28 | for arg in args: |
| 29 | if arg == '-' or arg == '': |
| 30 | sts = dofile(sys.stdin) or sts |
| 31 | elif os.path.isdir(arg): |
| 32 | sts = mh(arg) or sts |
| 33 | elif os.path.isfile(arg): |
| 34 | try: |
| 35 | f = open(arg) |
Guido van Rossum | b940e11 | 2007-01-10 16:19:56 +0000 | [diff] [blame] | 36 | except IOError as msg: |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 37 | sys.stderr.write('%s: %s\n' % (arg, msg)) |
| 38 | sts = 1 |
| 39 | continue |
| 40 | sts = dofile(f) or sts |
| 41 | f.close() |
| 42 | else: |
| 43 | sys.stderr.write('%s: not found\n' % arg) |
| 44 | sts = 1 |
| 45 | if sts: |
| 46 | sys.exit(sts) |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 47 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 48 | numeric = re.compile('[1-9][0-9]*') |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 49 | |
| 50 | def mh(dir): |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 51 | sts = 0 |
| 52 | msgs = os.listdir(dir) |
| 53 | for msg in msgs: |
| 54 | if numeric.match(msg) != len(msg): |
| 55 | continue |
| 56 | fn = os.path.join(dir, msg) |
| 57 | try: |
| 58 | f = open(fn) |
Guido van Rossum | b940e11 | 2007-01-10 16:19:56 +0000 | [diff] [blame] | 59 | except IOError as msg: |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 60 | sys.stderr.write('%s: %s\n' % (fn, msg)) |
| 61 | sts = 1 |
| 62 | continue |
| 63 | sts = message(f) or sts |
| 64 | return sts |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 65 | |
| 66 | def mmdf(f): |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 67 | sts = 0 |
| 68 | while 1: |
| 69 | line = f.readline() |
| 70 | if not line: |
| 71 | break |
| 72 | if line == '\1\1\1\1\n': |
| 73 | sts = message(f, line) or sts |
| 74 | else: |
| 75 | sys.stderr.write( |
| 76 | 'Bad line in MMFD mailbox: %r\n' % (line,)) |
| 77 | return sts |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 78 | |
Guido van Rossum | ee24811 | 1995-01-12 12:40:48 +0000 | [diff] [blame] | 79 | counter = 0 # for generating unique Message-ID headers |
| 80 | |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 81 | def message(f, delimiter = ''): |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 82 | sts = 0 |
| 83 | # Parse RFC822 header |
| 84 | m = rfc822.Message(f) |
| 85 | # Write unix header line |
| 86 | fullname, email = m.getaddr('From') |
| 87 | tt = m.getdate('Date') |
| 88 | if tt: |
| 89 | t = time.mktime(tt) |
| 90 | else: |
| 91 | sys.stderr.write( |
Barry Warsaw | 820c120 | 2008-06-12 04:06:45 +0000 | [diff] [blame] | 92 | 'Unparseable date: %r\n' % (m.get('Date'),)) |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 93 | t = os.fstat(f.fileno())[stat.ST_MTIME] |
Collin Winter | 6f2df4d | 2007-07-17 20:59:35 +0000 | [diff] [blame] | 94 | print('From', email, time.ctime(t)) |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 95 | # Copy RFC822 header |
| 96 | for line in m.headers: |
Collin Winter | 6f2df4d | 2007-07-17 20:59:35 +0000 | [diff] [blame] | 97 | print(line, end=' ') |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 98 | # Invent Message-ID header if none is present |
Collin Winter | 6f2df4d | 2007-07-17 20:59:35 +0000 | [diff] [blame] | 99 | if 'message-id' not in m: |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 100 | global counter |
| 101 | counter = counter + 1 |
| 102 | msgid = "<%s.%d>" % (hex(t), counter) |
| 103 | sys.stderr.write("Adding Message-ID %s (From %s)\n" % |
| 104 | (msgid, email)) |
Collin Winter | 6f2df4d | 2007-07-17 20:59:35 +0000 | [diff] [blame] | 105 | print("Message-ID:", msgid) |
| 106 | print() |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 107 | # Copy body |
| 108 | while 1: |
| 109 | line = f.readline() |
| 110 | if line == delimiter: |
| 111 | break |
| 112 | if not line: |
| 113 | sys.stderr.write('Unexpected EOF in message\n') |
| 114 | sts = 1 |
| 115 | break |
| 116 | if line[:5] == 'From ': |
| 117 | line = '>' + line |
Collin Winter | 6f2df4d | 2007-07-17 20:59:35 +0000 | [diff] [blame] | 118 | print(line, end=' ') |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 119 | # Print trailing newline |
Collin Winter | 6f2df4d | 2007-07-17 20:59:35 +0000 | [diff] [blame] | 120 | print() |
Tim Peters | e6ddc8b | 2004-07-18 05:56:09 +0000 | [diff] [blame] | 121 | return sts |
Guido van Rossum | 524f146 | 1994-09-05 11:51:33 +0000 | [diff] [blame] | 122 | |
Johannes Gijsbers | 7a8c43e | 2004-09-11 16:34:35 +0000 | [diff] [blame] | 123 | if __name__ == "__main__": |
| 124 | main() |